A monthly average is a useful summary and a poor alarm. When turbidity is reported as a monthly mean, a few minutes of elevated readings after a backwash or a change in raw water can disappear into the arithmetic. The figure on the report looks steady. The filter record underneath it may not be.
The mechanism is straightforward. Turbidity measures suspended particles, and in a treatment train it responds quickly to events an operator may not see for hours: a raw water surge after rain, a filter that does not settle properly after backwash, or a coagulant dose that drifts for one shift. Each excursion may last only minutes. A mean taken across thirty days gives those minutes very little weight against a large number of stable readings, so the average can barely move while particles still pass into the finished water for a measurable period.
Where the Peak Lives
Peaks sit in at least three records, and each answers a different question. Raw water turbidity shows what arrived at the plant. Individual filter effluent turbidity shows whether a given unit is performing. Combined or finished water turbidity shows what enters the distribution system. A plant-wide average can look reassuring while one filter has run high through an afternoon, because the other filters carry the figure down.
The practical step is to ask for the maximum, the time each maximum occurred, and the duration of any period above a level the plant has chosen to watch. Many plants already log at short intervals, so the underlying data often exists. The open question is whether the monthly report draws from that data, and whether anyone reads the trend before the report is signed. Regulatory reporting relies on defined methods, sampling frequencies, and limits that depend on the rule that applies to a given system and treatment type. Confirm those requirements against the permit and the state program before changing what gets reported or how alarms are set.
What Remote Feeds Change
Remote monitoring makes the peak easier to capture and easier to overlook. A feed can deliver readings every few minutes to a screen that no one watches continuously, and the logic that decides what is flagged is usually configured once and rarely reviewed. Vendor activity in the sector is current. Badger Meter acquires Melbourne-based remote monitoring company, as Water Finance and Management reported, describing an announcement made on Nov. 3 that Badger Meter had acquired 37South, a "utility technology and remote monitoring company" based in Melbourne, Australia. The deal does not change how a plant should read its own filter records, but it is a reminder that sensor data often passes through systems run by someone other than the utility before anyone at the plant sees it.
That distance matters for the record. The plant's own file should hold the maximums and timestamps it would need to reconstruct a filter event. A short export at a set interval is enough to rebuild a peak after the fact. Without it, a question about a filter event may rest on a monthly average that never showed the event at all.
What to Record
A workable log is short. For each filter, record the maximum effluent turbidity per day, the time of that maximum, the minutes above the level the plant has set for review, and the backwash that preceded it. For raw water, note rainfall or source changes on the same line, since the cause often sits in the date rather than the reading. Keep the averages too, but treat them as a second column, not the headline.
This approach will not settle every question. A short peak may reflect a real treatment problem, a sensor fault, or a sampling artifact, and the log alone does not say which. Tracing the cause takes the filter history, the calibration record, and the operator's notes. What the log does is preserve the event so those checks remain possible after the averages have closed the month.